MATERIALS AND METHODS
The primary study used a 3 session, double-blind, crossover design with administration of an opioid antagonist (naloxone), an opioid analgesic (morphine), or saline placebo in each session. Results reported below were cross-sectional, reflecting measures obtained at the initial baseline screening visit (before any drug administration or laboratory procedures). Identical data collection procedures were employed in a closely coordinated manner at 2 sites (Vanderbilt University Medical Center and Rush University Medical Center).
Participants were 87 individuals with CLBP who were recruited through on-line advertisements on the Vanderbilt University e-mail recruitment system, the Rush University Pain Clinic, advertisements in local print media, or posted flyers. Inclusion criteria were: (1) age between 18 and 55 years, (2) no self-reported history of cardiovascular disease, hypertension, liver, kidney disorders, diabetes, or seizure disorder, (3) no self-reported history of posttraumatic stress disorder, bipolar disorder, psychotic disorder, alcohol or drug dependence, (4) no use of antihypertensive medications, (5) no daily use of opioid analgesics, (6) chronic daily low back pain of at least 3 months duration, and (7) an average past month severity of at least 3 on a 0-10 verbal numeric pain intensity scale. Individuals with chronic pain related to malignancy, autoimmune disorders, or fibromyalgia were excluded. Potential participants who were pregnant (determined by urine pregnancy screens) were excluded to avoid fetal exposure to naloxone and morphine administered as part of the primary study. In total, 13 participants reported occasional use of as-needed opioid analgesics, but none used any opioids in the 3 days preceding the study procedures described below (confirmed by urine opiate screen). Three participants reported use of antidepressant medications. Participants were compensated $375 for their time upon completion of all study procedures. See Table 1 for sample characteristics.
Sleep disturbance was assessed using the PROMIS sleep disturbance—short form 8a,12 an 8 item instrument assessing sleep quality, sleep depth, and restoration associated with sleep. A score of 50 represents the average score in the normative sample, with higher scores indicating greater sleep disturbance. The PROMIS sleep disturbance measure has been shown to have strong psychometric properties, and demonstrates better measurement precision than other commonly used self-report sleep measures.12
Chronic Pain Intensity Measures
Chronic pain intensity was assessed using the McGill Pain Questionnaire-Short Form (MPQ).13 Instructions were modified to refer specifically to the low back pain experienced “over the past month.” The MPQ is a well-validated measure that allows separate assessment of the sensory (MPQ-sensory) and affective (MPQ-affective) qualities of pain.13 The MPQ also includes a visual analog pain intensity scale (VAS), anchored with “no pain” and “worst possible pain,” which was used to assess overall chronic pain intensity in the past month.
The Beck Depression Inventory (BDI)14 was used to assess recent depressive symptoms. Higher scores on the BDI indicate greater depressive symptoms, with scores of 10 or more indicating at least mild depressive symptoms. The trait form of the State-Trait Anxiety Inventory (STAI)15 was used to measure the tendency to experience anxiety symptoms. Higher scores on the STAI indicate greater tendency to experience anxiety symptoms. Both the BDI and STAI are commonly used self-report measures which have well-established psychometric properties.
To provide a complementary measure of affect, the Positive and Negative Affect Schedule (PANAS)16 was used. The PANAS contains brief negative affect (NA) and positive affect (PA) subscales, each consisting of 10 items. Higher scores indicate greater global negative and positive affect, respectively. Both PANAS subscales have shown good psychometric characteristics.16 For the current study, only the PANAS-PA scale was used, given the high degree of overlap between the PANAS-NA subscale and the BDI and STAI.
Pain catastrophizing was assessed using the Pain Catastrophizing Scale (PCS).17 The PCS is a 13-item measure which assesses the degree to which individuals have negative thoughts and feelings when experiencing pain. The PCS has been shown to have strong psychometric characteristics.17 Higher scores indicate greater pain catastrophizing.
To assess functional status, we administered four functional subscales of the Rand-36 Health Survey.18 The Rand-36 consists of the same 36 items included in the SF-36 questionnaire frequently employed in medical outcomes research.19 The Rand-36, however, employs a simpler unweighted scoring system.18 Subscales administered were: Physical Functioning, Social Functioning, Role Limitations because of Physical Health, and Role Limitations because of Emotional Problems. Subscale scores range from 0 to 100, with lower scores on each of these subscales indicating greater functional impairment.
All procedures were conducted at the Vanderbilt General Clinical Research Center or a dedicated research room at the Rush University Pain Center. All procedures were approved by the Institutional Review Boards at the respective institutions. After providing informed consent, participants completed a packet of questionnaires, including information regarding demographics, as well as the chronic pain, psychosocial status, and functional measures described above.
All analyses were conducted using IBM SPSS for Windows Version 24 (SPSS Inc., Chicago, IL). First, Pearson zero-order correlations were generated to examine relationships among the targeted variables. Next, evaluation of hypothesized statistical mediation models was carried out. Primary analyses were conducted in 2 sets, reflecting different hypothetical causal paths. The first (see conceptual model in Fig. 1) tested models in which the influence of sleep disturbance on chronic pain intensity was because of indirect effects of each of the psychosocial variables. The second (see conceptual model in Fig. 2) tested models in which the influence of sleep disturbance on functional measures was because of indirect effects of each of the chronic pain intensity measures. As described by Preacher and Hayes20 2 criteria were necessary to demonstrate mediation. Using the sleep disturbance and chronic pain intensity association as an example, to support mediation there needed to be: (1) a significant association between PROMIS sleep disturbance scores and chronic pain intensity measures (ie, there was an effect to be mediated), and (2) a significant indirect effect of PROMIS sleep disturbance scores on chronic pain intensity by the psychosocial measure tested. Where criterion 1 was met, the approach of Preacher and Hayes20 was used to test the significance of the indirect effect. To provide a conservative test for significance of direct effects of sleep disturbance on pain intensity and functional outcomes independent of all mediators examined, secondary analyses tested 2 multiple mediation models. The first was a model in which the effects of sleep disturbance on chronic pain intensity were mediated simultaneously by BDI, STAI, PANAS-PA, and CATS. The second was a model in which the effects of sleep disturbance on functional measures were mediated simultaneously by MPQ-Sensory, MPQ-Affective, and VAS intensity measures. Given the cross-sectional nature of the data, results of mediation analyses reflected statistical mediation only, and were unable to evaluate temporal precedence. Phrases used below describing effects “conveyed by” or “through” mediators are used to maximize readability, but should not be interpreted as indicating causality.
Direct and indirect effects were examined for all models above using bootstrapped mediation analysis20 by custom SPSS dialogue (the PROCESS Procedure; http://www.processmacro.org/index.html). Bootstrapping is a nonparametric re-sampling procedure that does not assume normality for the indirect effect, or a*b path coefficient. This nonparametric feature differs from the “causal steps” approach to indirect effects (mediation) testing.21 Bootstrapping involves random and repeated sub-sampling of the data to derive an estimate of the a*b (indirect effect) path coefficient and a 95% confidence interval around the bootstrapped indirect effect parameter. If the 95% confidence interval for the indirect effect generated by the model does not include zero, then the hypothesized indirect (mediation) effect is significant at the P<0.05 level. Mediation analyses specified model 4 and k=10,000 bootstraps. Percentile bootstrapping was used to derive path coefficients and corresponding 95% confidence intervals for indirect effects. On the basis of the empirically derived guidelines provided by Fritz and MacKinnon,22 the study sample size (n=87) was sufficient to detect indirect (mediated) effects with >0.80 power, assuming at least a moderate effects size for the association between sleep and the targeted mediator and the association between the mediator and the pain or functional outcome. Such effect sizes would likely be required for these mediated effects to be clinically meaningful.
Table 1 indicates that the sample was comprised of individuals who had experienced CLBP for an average of almost 10 years, and who on average reported moderate pain intensity on the VAS measure. The mean level of sleep disturbance in the sample was comparable to the mean observed in the PROMIS normative sample (ie, T=50), although sleep disturbance levels were elevated at least one standard deviation above this mean in 30.3% of study participants (ie, T score ≥60). Participants did not, on average, report a high level of functional impairment (Rand-36) or psychosocial distress (BDI, STAI), consistent with the nature of the sample (ie, not using opioid analgesics regularly, the majority not being treated at a tertiary care chronic pain clinic).
Zero-order correlations between sleep disturbance and chronic pain intensity measures, and measures of psychosocial status and function, are summarized in Table 2. Higher levels of sleep disturbance on the PROMIS measure were correlated positively with all 3 measures of chronic back pain intensity (medium to large effect sizes). Higher sleep disturbance scores were also associated with greater depression (BDI), anxiety (STAI), and catastrophizing (CATS), lower global positive affect (PANAS-PA), and worse functioning on all 4 Rand-36 subscales examined. Hypothesized mediators in both proposed models were associated as expected with targeted outcomes. Elevated depression (BDI), anxiety (STAI), catastrophizing (CATS), and lower positive affect (PANAS-PA) were all significantly correlated in the expected direction with chronic pain intensity outcomes (MPQ-sensory, MPQ-affective, VAS intensity). Greater chronic pain intensity on all 3 of these measures was also significantly associated with worse function on all 4 Rand-36 subscales.
Tests of Mediation
The results of zero-order correlation analyses indicated that preconditions for testing mediation were met for all associations between sleep disturbance and the chronic pain and functional outcomes. Results of the first series of mediation analyses testing whether the association between sleep disturbance and chronic pain intensity was mediated by psychosocial status (with each potential mediator tested in a separate model) are summarized in Table 3. Significant indirect effects of sleep disturbance on all 3 chronic pain intensity outcomes by depression levels (BDI) were observed (ie, the bootstrap confidence intervals did not include zero). Similarly, significant indirect (mediated) effects of sleep disturbance on both MPQ-Affective and VAS pain intensity ratings by catastrophizing (CATS) were also noted. The mediation model focused on the MPQ-Affective pain rating outcome (ie, pain-related suffering) was the only one to reveal significant indirect effects of sleep disturbance conveyed statistically through elevated anxiety (STAI) and reduced positive affect levels (PANAS-PA). Direct effects of sleep disturbance on chronic pain intensity outcomes were significant for 10 of the 12 models tested, indicating partial rather than full mediation by the psychosocial variables evaluated. That is, greater sleep disturbance is associated with greater chronic pain intensity both through the mediating effects of psychosocial distress and catastrophizing, as well as through associations with pain intensity that are independent of psychosocial factors.
The second series of mediation analyses (Table 4) evaluated the hypothesis that the effects of sleep disturbance on functional outcomes were mediated by chronic pain intensity (with each potential mediator tested in a separate model). Significant indirect effects of sleep disturbance on the 4 functional (Rand-36) outcomes were observed for all 3 hypothesized chronic pain intensity mediators. Indirect (mediated) effects were most consistently observed for the MPQ-affective measure. These findings support a model in which the influence of greater sleep disturbance on higher functional impairment is conveyed in part by the indirect effects of sleep disturbance on higher chronic pain intensity. Nonetheless, significant direct effects for all 12 models tested indicate partial rather than full mediation. In other words, greater sleep disturbance is associated with elevated functional impairment in part because of the association between sleep disturbance and increased chronic pain intensity, but also through direct influences of sleep disturbance independent of pain intensity.
In secondary analyses, we conducted multiple mediation tests to provide a conservative test of the direct effects of sleep disturbance on both chronic pain intensity and functional outcomes independent of the multiple mediators (entered jointly) identified in primary analyses. Results indicated significant direct effects of sleep disturbance on MPQ-sensory ratings (t=2.41, P=0.0183) and VAS intensity ratings (t=3.46, P=0.009) independent of the multiple mediators (BDI, STAI, PANAS-PA, and CATS). In addition, the total indirect (mediated) effect by the multiple mediators was significant for VAS intensity (95% CI for total indirect effect: 0.030, 0.6797), but not for MPQ-sensory ratings (95% CI for total indirect effect: −0.003, 0.2667). For the MPQ-affective outcome, the direct effect independent of the multiple mediators was not significant (t=1.066, P=0.29), but there was a significant indirect (mediated) effect by the multiple mediators (95% CI for total indirect effect: 0.0287, 0.1662).
For the functional outcomes, results of these secondary analyses indicated significant direct effects of sleep disturbance on Rand-36 functional limitations (t=−5.374, P<0.001), Rand-36 Social Functioning (t=−2.401, P=0.0186), Rand-36 physical role limitations (t=−2.271, P=0.026), and Rand-36 emotional role limitations (t=−2.69, P=0.0086) independent of the multiple mediators (MPQ-sensory, MPQ-affective, VAS intensity). In addition, the total indirect (mediated) effects by the multiple mediators were significant for Rand-36 physical functioning (95% CI for total indirect effect: −0.9722, −0.168), Rand-36 social functioning (95% CI for total indirect effect: −1.0744, −0.1995), and Rand-36 physical role limitations (95% CI for total indirect effect: −1.456, −0.1738), but not for Rand-36 emotional role limitations (95% CI for total indirect effect: −1.1668, 0.0126).
Evidence strongly suggests that sleep disturbance, chronic pain intensity, and functional impairment are related, either directly and/or by intermediate factors. The present study examined these direct and indirect (mediated) pathways through which sleep disturbance might be related to chronic pain intensity and functional impairment. Findings support both direct and indirect associations. Greater sleep disturbance was associated with greater pain intensity, greater pain-related dysfunction, and greater psychosocial distress and catastrophizing.3,9 Given the pattern of observed associations, we tested 2 series of mediation models. The first model tested whether previously reported links between sleep disturbance and greater pain intensity were mediated by negative psychosocial status associated with sleep disturbance. Results as a whole suggested that positive associations between sleep disturbance and chronic pain intensity were all conveyed statistically in part by the indirect effects of depressive symptoms, anxiety, catastrophizing, and positive affect (inverse) associated with sleep disturbance. This pattern showed consistency particularly for depression, and to a lesser degree catastrophizing, across multiple indices of chronic pain intensity. The most consistent mediation effects for psychosocial status were noted for affective chronic pain ratings (MPQ-affective). Overall, 7 of 12 mediation models that were evaluated indicated significant indirect (mediated) effects of sleep disturbance by psychosocial status. That is, sleep disturbance may exert deleterious effects on pain by increasing negative affect and catastrophizing and by reducing positive affect, which may in turn lead to increased pain intensity. This result is consistent with previous reports of psychosocial distress mediating the relationship between sleep and pain.10,23 This result also adds to recent literature suggesting that psychological therapies, such as cognitive-behavioral therapy, may improve chronic pain intensity in part by reducing psychosocial distress.24 However, the fact that 10 of the 12 mediation models tested also revealed significant direct effects indicated that psychosocial status partially rather than fully mediated most of the observed associations between sleep disturbance and chronic pain intensity. Thus, sleep disturbance may worsen pain intensity by other mechanisms not tapped by psychosocial factors, such as increased inflammation.25
The second series of mediation models evaluated whether associations between sleep disturbance and greater functional impairment were mediated in part through the indirect effects of elevated pain intensity. Nine of the 12 mediation models tested across the 4 functional measures revealed significant indirect effects. Specifically, the association between sleep disturbance and impaired physical and social functioning, and role limitations because of physical and emotional factors, were all at least partly accounted for by the mediating effects of greater pain intensity. As for the psychosocial mediation models, partial rather than full mediation by pain intensity was suggested.
The overall pattern of study findings suggests that sleep disturbance may negatively impact both chronic pain intensity and functional status independent of a number of plausible theory-driven mediators. These results are consistent with previous reports from more heterogeneous chronic pain samples5,6 indicating that sleep disturbance can have direct effects on chronic pain intensity and separate direct effects on function. This interpretation is supported by the significant direct effects observed in multiple mediation models (in secondary analyses). Adding to this prior work, however, these multiple mediation models also suggest that the negative effects of sleep disturbance on chronic pain intensity may derive in part from the adverse effects of sleep disturbance on psychosocial status. Moreover, these models further suggest that sleep disturbance may adversely impact functional status in chronic back pain patients in part because of associations between sleep disturbance and elevated chronic pain intensity. This latter finding is in contrast to previous work suggesting that when sleep duration and sleep quality were considered jointly, neither predicted results of behavioral functional tests once the influence of pain intensity and catastrophizing were controlled.5 Differing results for functional measures between our work and this prior work may be because of the different sleep measures employed, different functional measures used (broad self-reported function versus specific behavioral function tests), as well as different analytical approaches (mediation analysis versus multiple regression). Moreover, differences could have been due in part to presence of medication confounds in prior work. Because of inclusion criteria of the primary study, no patients in the current study were using daily opioids and none had recently used as-needed opioids (confirmed by urine drug screen).
These results further highlight the potential benefits of screening for and treating sleep disturbance within the context of chronic pain management. Sleep disturbance can result from a variety of sleep disorders such as sleep disordering breathing (sleep apnea) and insomnia, and risk for these can be assessed with various questionnaires.26,27 Our results, along with others, raise the possibility that treating chronic pain patients for these causes of sleep disturbance (ie, continuous positive airway pressure [CPAP], cognitive-behavioral therapy for insomnia [CBT-I]) could lead to improvements in chronic pain intensity and function, both directly and by improvements in psychosocial status. Indeed, CPAP has been shown to reduce bodily pain, reduce pain sensitivity, and improve function in patients with sleep apnea,28,29 findings consistent with the significant direct effects of sleep disturbance on pain and function reported in the current work. In addition, CBT-I has been shown to improve sleep and also reduce pain intensity and pain interference in chronic pain patients.30–32 Only one of these studies examined associations between the degree of improvement in sleep and pain outcomes, and found that the improvement in sleep significantly predicted subsequent decreases in pain.32
Some limitations of this study should be acknowledged. The first is that the design was cross-sectional, and therefore the results reported describe statistical mediation only. A longitudinal design assessing chronic pain, sleep disturbance, psychosocial status, and function at multiple points over time would be required to demonstrate true causal mediation. This study also conducted a relatively large number of mediation tests, which could potentially have inflated the type I error rate. Risks that the mediation results reported above are due solely to Type I error are mitigated to some extent by the fact that significant results were largely consistent across the 3 chronic pain outcomes and the 4 functional measures evaluated (16/24 mediation tests were significant). Nonetheless, replication of these findings would be desirable. Finally, sleep disturbance was self-reported and there was no objective assessment of sleep such as can be derived from wrist actigraphy.
In summary, results of this study add to an emerging literature that highlights the potential negative effects of sleep disturbance on chronic pain intensity and function in patients with chronic pain. Here, we contribute to the literature in 2 ways. First, we showed that the detrimental effects of sleep disturbance on pain and function can be explained partly through the detrimental effects sleep disturbance may have on psychosocial factors, such as depressed mood. Sleep disturbance may worsen mood which in turn may worsen pain. Second, we showed that some of the effects of sleep disturbance on pain and function cannot be explained by ill effects of sleep disturbance on psychosocial factors. Thus, other mechanisms by which poor sleep affects pain and function remain to be determined.
1. Jegan NR, Brugger M, Viniol A, et al. Psychological risk and protective factors for disability in chronic low back pain
- a longitudinal analysis in primary care. BMC Musculoskelet Disord. 2017;18:114.
2. Bahouq H, Allali F, Rkain H, et al. Prevalence and severity of insomnia in chronic low back pain
patients. Rheumatol Int. 2013;33:1277–1281.
3. Tang NK, Wright KJ, Salkovskis PM. Prevalence and correlates of clinical insomnia co-occurring with chronic back pain
. J Sleep
4. Finan PH, Goodin BR, Smith MT. The association of sleep
: an update and a path forward. J Pain
5. Roberts MB, Drummond PD. Sleep
problems are associated with chronic pain
over and above mutual associations with depression and catastrophizing. Clin J Pain
6. McCracken LM, Iverson GL. Disrupted sleep
patterns and daily functioning in patients with chronic pain
Res Manag. 2002;7:75–79.
7. Finan PH, Quartana PJ, Remeniuk B, et al. Partial sleep
deprivation attenuates the positive affective system: effects across multiple measurement modalities. Sleep
8. Haack M, Mullington JM. Sustained sleep
restriction reduces emotional and physical well-being. Pain
9. Gerhart JI, Burns JW, Post KM, et al. Relationships between sleep
quality and pain
-related factors for people with chronic low back pain
: tests of reciprocal and time of day effects. Ann Behav Med. 2017;51:365–375.
10. O’Brien EM, Waxenberg LB, Atchison JW, et al. Negative mood mediates the effect of poor sleep
among chronic pain
patients. Clin J Pain
11. Naughton F, Ashworth P, Skevington SM. Does sleep
quality predict pain
-related disability in chronic pain
patients? The mediating roles of depression and pain
12. Yu L, Buysse DJ, Germain A, et al. Development of short forms from the PROMIS sleep
disturbance and sleep
-related impairment item banks. Behav Sleep
13. Melzack R. The short-form McGill Pain
14. Beck AT, Ward CH, Mendelson M, et al. An inventory for measuring depression. Arch Gen Psychiatry. 1961;4:53–63.
15. Spielberger CD, Gorsuch RL, Lushene RE. The State-Trait Anxiety Inventory Manual. Palo Alto, CA: Consulting Psychologists Press; 1970.
16. Watson D, Clark LA, Tellegen A. Development and validation of brief measures of positive and negative affect: the PANAS scales. J Pers Soc Psychol. 1988;54:1063–1070.
17. Sullivan MJL, Bishop SR, Pivik J. The Pain
Catastrophizing Scale: development and validation. Psychol Assess. 1995;7:524–532.
18. Hays RD, Sherbourne CD, Mazel RM. The RAND 36-item health survey 1.0. Health Econ. 1993;2:217–227.
19. Ware JE Jr, Sherbourne CD. The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and item selection. Med Care. 1992;30:473–483.
20. Preacher KJ, Hayes AF. SPSS and SAS procedures for estimating indirect effects in simple mediation
models. Behav Res Methods Instrum Comput. 2004;36:717–731.
21. Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol. 1986;51:1173–1182.
22. Fritz MS, Mackinnon DP. Required sample size to detect the mediated effect. Psychol Sci. 2007;18:233–239.
23. Koffel E, Krebs EE, Arbisi PA, et al. The unhappy triad: pain
complaints, and internalizing symptoms. Clin Psychol Sci. 2016;4:96–106.
24. Dura-Ferrandis E, Ferrando-Garcia M, Galdon-Garrido MJ, et al. Confirming the mechanisms behind cognitive-behavioural therapy effectiveness in chronic pain
using structural equation modeling in a sample of patients with temporomandibular disorders. Clin Psychol Psychother. 2017;24:1377–1383.
25. Haack M, Sanchez E, Mullington JM. Elevated inflammatory markers in response to prolonged sleep
restriction are associated with increased pain
experience in healthy volunteers. Sleep
26. Netzer NC, Stoohs RA, Netzer CM, et al. Using the Berlin questionnaire to identify patients at risk for the sleep
apnea syndrome. Ann Intern Med. 1999;131:485–491.
27. Bastien CH, Vallieres A, Morin CM. Validation of the Insomnia Severity Index as an outcome measure for insomnia research. Sleep
28. Zhao YY, Wang R, Gleason KJ, et al. Effect of continuous positive airway pressure treatment on health-related quality of life and sleepiness in high cardiovascular risk individuals with sleep
apnea: Best Apnea Interventions for Research (BestAIR) Trial. Sleep
29. Khalid I, Roehrs TA, Hudgel DW, et al. Continuous positive airway pressure in severe obstructive sleep
apnea reduces pain
30. Vitiello MV, Rybarczyk B, Von Korff M, et al. Cognitive behavioral therapy for insomnia improves sleep
and decreases pain
in older adults with co-morbid insomnia and osteoarthritis. J Clin Sleep
31. Jungquist CR, O’Brien C, Matteson-Rusby S, et al. The efficacy of cognitive-behavioral therapy for insomnia in patients with chronic pain
32. Smith MT, Finan PH, Buenaver LF, et al. Cognitive-behavioral therapy for insomnia in knee osteoarthritis: a randomized, double-blind, active placebo-controlled clinical trial. Arthritis Rheumatol. 2015;67:1221–1233.
Keywords:Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved.
chronic pain; function; pain; sleep; mediation